J Oceanogr (2011) 67:347–355 DOI 10.1007/s10872-011-0044-1

ORIGINAL ARTICLE

Different effects of tropical cyclones generated in the South Sea and the northwest Pacific on the summer circulation

Zheng Ling • Guihua Wang • Chunzai Wang • Zhi-Song Fan

Received: 2 October 2010 / Revised: 17 May 2011 / Accepted: 24 May 2011 / Published online: 17 June 2011 Ó The Oceanographic Society of and Springer 2011

Abstract Tropical cyclones (TCs) that affect the South southwesterly and northeasterly winds prevail in the SCS in China Sea (SCS) can be generated in either the SCS or the summer and winter, respectively. The southwesterly wind northwestern Pacific (NWP). Using satellite measurements, usually occurs around mid-May in the southern and central the Sverdrup theory and a 1.5-layer nonlinear reduced SCS, then expands to the entire SCS in June, and matures gravity model, the present paper investigates the effects of from July to August. The strong northeasterly winter SCS and NWP TCs on the summer SCS upper layer ocean monsoon first appears over its northern part in September, circulation. Both SCS and NWP TCs enhance the summer reaching its central part in October and covering the entire mean circulation pattern of the cyclonic gyre in the SCS in November, and then gradually diminishing in April. northern SCS and the anti-cyclonic gyre in the southern The SCS is also an ocean basin that is often affected by SCS. However, the effect of SCS TCs is much larger than tropical cyclones (TCs) generated either over the north- that of NWP TCs, although the number of SCS TCs is western Pacific (NWP) or the SCS (Wang et al. 2007). As smaller than NWP TCs. This is because the SCS TCs- in many studies, we define TC as a tropical mesoscale induced wind stress curl pattern is favorable for enhancing cyclonic weather system with its strength of a tropical the summer SCS mean circulation. storm or stronger (i.e., typhoon). Our TC dataset shows that, among the summer (winter) average of 5.3 (3) TCs Keywords Upper layer circulation Tropical cyclones passing through the SCS, about 1.5 (0.25) TCs are origi- South China Sea The northwestern Pacific nated or generated within the SCS and the remaining TCs are originated from the NWP. Since large-scale upper layer ocean circulation in the 1 Introduction SCS is mainly driven by the surface wind, both the mon- soon-related and TC-induced winds should play a role in The South China Sea (SCS) is a semi-enclosed ocean basin SCS circulation. How the SCS upper layer responds to the temporally forced by the pronounced monsoon wind. The monsoon has been well studied (Liu et al. 2001; Wang et al. 2006). It has been shown that a winter basin-wide cyclonic gyre with two cyclonic cells is forced by the Z. Ling Z.-S. Fan College of Physical and Environmental Oceanography, winter monsoon, while a weakened cyclonic gyre north of Ocean University of China, Qingdao, China about 12°N and a strong anti-cyclonic gyre south of about 12°N are produced by the summer monsoon. Associated & Z. Ling G. Wang ( ) with these gyres are strong western boundary currents. In State Key Laboratory of Satellite Ocean Environment Dynamics, Second Institute of Oceanography, SOA, Hangzhou, China winter, a southward jet flows along the entire western e-mail: [email protected] boundary (Wrytki 1961; Liu et al. 2001; Wang et al. 2008). In summer, there is a northward jet flowing along the C. Wang western boundary in the southern SCS (Xu et al. 1982). Physical Oceanography Division, NOAA/Atlantic Oceanographic and Meteorological Laboratory, The jet apparently veers eastward off central near Miami, FL, USA 12°N (Xu et al. 1982; Xie et al. 2003; Wang et al. 2003, 123 348 Z. Ling et al.

2006). To the north, there is a narrow southwestern speed. We choose all TCs that could have an impact on the jet along the continental slope in the northwest SCS (Qu SCS, defined for this purpose as the area of 0–22.5°N, 2000; Metzger and Hurlburt 2001;Su2004). 98.5–120.5°E. However, there are few studies that have examined the The QuikSCAT wind dataset we used is from the Jet SCS response to TC-induced wind. Chu et al. (1999) used Propulsion Laboratory. The spatial resolution of Quik- the Princeton Ocean Model to investigate the SCS response SCAT wind dataset is 0.25° latitude by 0.25° longitude and to TC ‘Ernie’ and studied the sea surface temperature the time resolution is 1 day. To evaluate the QuikSCAT cooling to the right of TC track and a very strong divergent wind data quality, we compared the QuikSCAT wind in the upper layer current. Lin et al. (2003) combined several SCS with the observed winds from two islands during May datasets to show the effects of TCs on chlorophyll and 2005–December 2006. The root mean square errors marine life. Recently, Wang et al. (2009) investigated the between the QuikSCAT and island observations for wind seasonal variability of SCS ocean circulation induced by speed and wind direction are 1.65 m/s and 27.8°, respec- TCs using satellite QuikSCAT wind field data and a tively. These values are consistent with the results of Liu reduced gravity model. They found that TCs can affect and He (2003) who compared the QuikSCAT wind in the both large-scale and mesoscale SCS ocean circulation. As SCS with the observed wind on seven islands. As dem- stated earlier, TCs that affect the SCS can be formed in onstrated in previous studies, the QuikSCAT wind over the either the SCS or the NWP. However, whether TCs gen- SCS is useful to simulate SCS circulation (Xie et al. 2003; erated in the SCS and the NWP have different effects on Wang et al. 2006, 2008), although the QuikSCAT wind SCS circulation is unknown. The purpose of the present shows some differences from island observed wind. paper is to answer and address this question. Because there To examine the effect of TCs on SCS ocean circulation, are few SCS TCs during the northeasterly monsoon period, we reconstructed four sets of wind products from the as shown in Table 1, NWP TCs dominate the influence of QuikSCAT wind dataset as follows: TCs on the mean SCS circulation with an enhancement of A. The original QuikSCAT wind data. the basin-wide cyclonic flow in this period. Thus, this study B. The wind data excluding all TCs. will focus on different effects of TCs generated in the SCS C. The wind data excluding TCs generated over the and the NWP on the summer SCS circulation during the NWP. southwesterly monsoon period. D. The wind data excluding TCs generated over the SCS. We will first use the QuikSCAT wind, one of the best wind products available, to estimate the wind stress curls Hereinafter, we will use A, B, C and D to denote the induced by SCS and NWP generated TCs. A simple model four wind forcings, respectively. Note that a simple linear of the Sverdrup theory and a 1.5-layer reduced gravity interpolation in time is used to reconstruct the wind forcing model are then used to investigate the effects of TCs for those periods during which we exclude TCs. For generated in the SCS and NWP on SCS ocean circulation. example, if a TC occurs from 6 to 8 June, we remove the 3 days’ original wind fields first, and then use the wind field on 5 and 9 June to obtain the wind field for the 2 Data removed 3 days with the linear interpolation method. The simple method is very effective in excluding TCs, but The TC dataset is from the Japanese Meteorological leaving the background wind field (the monsoon wind). Agency. The cover period of the dataset is from January A multiple altimeter dataset from the US Naval 2000 to December 2007. The starting year of 2000 is Research Laboratory, derived from TOPEX/POSEIDON, chosen because it is the beginning of a complete year for Jason, ERS and Geosat Follow-On (GFO), is used to val- the QuikSCAT wind data. The TC dataset includes idate our model results. The spatial resolution of the dataset 6-hourly positions of each TC, its impact radius, and its is 0.25° 9 0.25° and the time resolution is daily within the associated minimum center pressure and maximum wind period from year 2000 to 2007.

Table 1 The number of tropical cyclones which affect the South China Sea from 2000 to 2007 Month Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

SCS tropical cyclones 0 0 0 0 4 3 4 6 5 1 0 1 NWP tropical cyclones 0 0 1 1 2 5 11 13 9 7 12 3 Total 0 0 1 1 6 8 15 19 14 8 12 4

123 Different effects of tropical cyclones generated in the South China Sea 349

3 Methods

3.1 Sverdrup theory

The Sverdrup theory is a primary balance for the large- scale wind-driven ocean circulation (Sverdrup 1947). Many studies have applied it to calculate the large-scale circulations in the Pacific, Atlantic and Indian Oceans. This simple model has also been successfully applied to study SCS circulation (Metzger and Hurlburt 1996; Shaw et al. 1999; Liu et al. 2001). The wind-driven transport stream- function (w) for flow integrated in the entire water column can be estimated by the Sverdrup theory as:

ZxE 1 wðxÞ¼ curlðsÞdx; qb x where q is the density of the water, b is the northward Fig. 1 The SCS and the 12-h locations of TCs in summer (triangles gradient of the planetary vorticity, and xE represents the represent NWP TCs and solid circles represent SCS TCs). The two ocean eastern boundary. The boundary condition is w = 0 isobaths are for 200 and 2,000 m, respectively. The open ocean boundaries are set at 18°N and 24.5°N lines, and the boundary along at xE. 126°E is set as a wall The curl of wind stress (s) in the Sverdrup relation can be calculated by the following formula: winds switched on gradually from rest to the wind distri- bution of 1 January 2000 over a 1-month period. The model osy osx curlðsÞ¼ : is then run with the wind forcing data of the whole year ox oy from 1 January to 31 December 2000. The model run is The zonal and meridional wind stress sx and sy in the repeated for 4 years until the ocean circulation reaches a above equation can be obtained as: quasi-equilibrium state. Then, the model is forced from 1 January 2000 to 31 December 2007 with the four wind sx ¼ q CDjjV u a forcings of A, B, C, and D as mentioned in Sect. 2. More sy ¼ qaCDjjV v details of this simple model can be found in Wang et al. 0:0012; jjV \11 (2006, 2008, 2009). The reduced gravity model output is C ¼ D ðÞ0:45 þ 0:069 jjV 0:001; jjV \11; daily, and we thus define the summer mean as the average of the daily dataset from 1 June to 31 August. where CD is the wind stress drag coefficient (Yin et al. To include the Kuroshio as a driving force in some 2007), qa is the density of air; |V| is the wind speed, u is the experiments, we open the Strait and specify a zonal wind, and v is the meridional wind. We use the Kuroshio input along the zonal section of 122–124.5°E monthly wind to drive the Sverdrup model. Thus, we define (18°N). The open boundary conditions are given, as done the summer mean is the average of the monthly dataset by Hurlburt and Thompson (1980). The idea of the open from June to August for the Sverdrup model output. boundary conditions is that: the profile of transport is prescribed for inflow at south and the normal flow at the 3.2 Reduced gravity model north port is self-determined with the integral constraint that the total transport of outflow matches the inflow, and A 1.5-layer nonlinear reduced gravity model is also the amplitude of the tangential component is set to zero employed to simulate the wind-driven upper ocean circu- one-half grid distance outside the physical domain at both lation. Our model boundary is along the 200-m isobaths south and north ports. A natural question is whether the set (Fig. 1). The reduced gravity is set to 0.03 m s-2, with the of open boundary conditions can produce more realistic lateral friction coefficient of 500 m2 s-1 and the initial simulations than the close boundary conditions. Figure 2a–c thermocline depth of 200 m. The model resolution is on a show the summer empirical orthogonal function (EOF) grid of 0.25° latitude by 0.25° longitude. The grid size is modes of MODAS sea surface height anomaly (SSHA) smaller than the first baroclinic Rossby radius of defor- and the simulated thermocline depth anomaly with two mation for the deep basin of the SCS, which is larger than experiments (including the Kuroshio and excluding the 50 km (Cai et al. 2008). The model is spun up with the Kuroshio). It should be noted that these modes are mainly 123 350 Z. Ling et al.

Fig. 2 EOF summer mode of a the SSHA (cm), b the modeled thermocline depth anomaly (m) with the Kuroshio, and c the modeled thermocline depth anomaly (m) without the Kuroshio; d their corresponding time series reflecting summer circulation although the principal com- have also been demonstrated by many other studies (e.g., ponents (PCs) show a robust seasonal variability. The Metzger and Hurlburt 1996; Liu et al. 2001; Wang et al. summer modes account for 49.7%, 43.9% and 40.9% of 2006, 2008). total variance for MODAS SSHA, the thermocline depth anomalies including Kuroshio and excluding the Kuroshio, respectively. The MODAS SSHA shows that there is an 4 Results anti-cyclonic anomaly south of 12°N, a cyclonic anomaly in the northwestern SCS and another anti-cyclonic anomaly 4.1 Sverdrup relationship northwest of Luzon Island. The patterns of the two simu- lated thermocline depth anomaly are generally similar to From 2000 to 2007, there were 42 TCs in summer that MODAS SSHA (a northern cyclonic gyre and a southern affected the SCS. Of these 42 TCs, 13 were generated over anti-cyclonic gyre driven by the monsoon). The correlation the SCS and the other 29 were formed in the NWP is 0.82 between the observed PC and the simulated PC (Table 1). Figure 1 shows the 12-h positions of the 42 TCs, excluding the Kuroshio. The correlation is 0.84 (slightly all of which were located to the north of 12°N. higher) for the case including the Kuroshio. This suggests In summer, the southwest monsoon induces positive that the open boundary condition for including the Kuro- (negative) wind stress curl in the northwestern (southeast- shio can slightly help by having more accurate simulations. ern) SCS (Fig. 3a). Figure 3b shows that the TCs in sum- The validity and effectiveness of such a simple model to mer induce a larger positive wind stress curl in the study the SCS circulation dynamics over the deep water northwestern SCS and negative wind stress curl in the

123 Different effects of tropical cyclones generated in the South China Sea 351

Fig. 3 TC-induced wind stress curls (10-8Nm-3) in summer. a The Fig. 4 Stream function (Sv 106 m3 s-1) calculated from the Sverdrup wind stress curl of the original wind forcing, b the wind stress curl theory in summer. a Mean stream function forced by the original difference between the original wind forcing and the wind forcing wind, b stream function difference between the original wind forcing without any TCs, c the wind stress curl difference between the wind and the wind forcing without any TCs, c stream function difference forcing only with NWP TCs and the wind forcing without any TCs, between the wind forcing only with NWP TCs and the wind forcing and d the wind stress curl difference between the wind forcing only without any TCs, and d stream function difference between the wind with SCS TCs and the wind forcing without any TCs. The positive forcing only with SCS TCs and the wind forcing without any TCs. values are shaded The positive values are shaded southeastern SCS. Although the general pattern of wind obtain the TC-induced Sverdrup stream function through stress curl induced by NWP TCs is similar to that induced the difference among large-scale circulations (Sverdrup by SCS TCs (Fig. 3c, d), there are some distinguished stream functions) forced by the wind field described in differences. The positive wind stress curl induced by SCS Sect. 2, not through TC wind driven directly. Figure 4b TCs is north of 15°N, whereas the positive wind stress curl indicates that TCs enhances both the cyclonic and anti- induced by NWP TCs retreats northwestward with a rela- cyclonic gyres in summer. As shown in Fig. 4c, d, both tively small band of the positive curl. Interestingly, for the NWP and SCS TCs can enhance the double gyre circula- case of the NWP TC-induced wind stress curl, a negative tion. However, the effect of SCS TCs is much larger than wind stress curl appears in the northwest of Luzon Island that of NWP TCs, although the number of NWP TCs in (Fig. 3c). summer is larger than SCS TCs. This is because the wind The summer SCS ocean circulation is dominated by stress curl pattern induced by SCS TCs is more favorable double gyres: a north cyclonic gyre and a south anti- for increasing the Sverdrup stream function than NWP TCs cyclonic gyre (Fig. 4a). Because TCs produce a positive (Fig. 3c, d). There are two reasons why SCS TCs produce and negative wind stress curl in the northwestern and more favorable wind stress curl pattern than NWP TCs. southeastern SCS, respectively (Fig. 3b), the TC-induced Firstly, the distribution of TCs shows that some NWP TCs Sverdrup stream function also shows a north cyclonic gyre are located in the northeastern SCS, whereas most SCS and a south anti-cyclonic gyre (Fig. 4b). Note that we TCs are in the central SCS (Fig. 1). When a NWP TC

123 352 Z. Ling et al.

Fig. 5 Thermocline depth (m) in summer from the model experiment Fig. 6 The same as Fig. 5, except for the model experiment without with the Kuroshio. a Mean thermocline depth forced by the original the Kuroshio. a The dashed (solid) contours are for the thermocline b wind, thermocline depth difference between the original wind depth smaller (larger) than 200 m. b–d The positive values are shaded forcing and the wind forcing without any TCs, c thermocline depth difference between the wind forcing only with NWP TCs and the wind forcing without any TCs, and d stream function difference we conduct a series of 1.5-layer reduced gravity model between the wind forcing only with SCS TCs and the wind forcing without any TCs. a The dashed (solid) contours are for the experiments to further investigate how NWP and SCS TCs thermocline depth smaller (larger) than 182 m. b–d The positive affect summer SCS circulation. We use the four wind values are shaded datasets of A, B, C and D discussed in Sect. 2 to drive the model with open boundary conditions. passes the northeastern SCS, it will induce a negative wind Figure 5a shows the modeled summer mean thermocline stress curl there, which is unfavorable for enhancing the depth over the SCS from 2000 to 2007. The summer mean cyclonic gyre in the northern SCS. Secondly, the time that circulation pattern includes a cyclonic gyre north of 12°N SCS TCs are centered over the SCS basin (109–121°E, and an anti-cyclonic gyre south of 12°N. Associated with 5°N–23°N) is longer than that of NWP TCs, at 43.5 and these two gyres, a dipole structure of oceanic eddy appears 37.5 days, respectively, during summer for the period from off central Vietnam with a cyclonic eddy north and an anti- 2000 to 2007. This indicates that SCS TCs may produce cyclonic eddy south (note that an eastward jet is between more positive wind stress curl in the SCS. the dipole eddies; Liu et al. 2001; Wang et al. 2010). The model run also demonstrates that TCs can enhance the 4.2 Reduced gravity model cyclonic gyre in the northern SCS and an anti-cyclonic gyre in the southern SCS (Fig. 5b). The pattern of the The Sverdrup theory assumes that the ocean is in near- summer thermocline depth induced by NWP and SCS TCs equilibrium with surface wind forcing and neglects the are similar (Fig. 5c, d), with both the cyclonic and anti- details of ocean adjustment. To consider the ocean cyclonic gyres being enhanced. However, as demonstrated adjustment processes and more complete ocean processes, by the Sverdrup theory in the last section, the enhancement

123 Different effects of tropical cyclones generated in the South China Sea 353 by SCS TCs is stronger than that by NWP TCs. As expected, the dipole structure of an anti-cyclonic eddy south of an eastward jet and a cyclonic eddy north is more intensified by SCS TCs than NWP TCs. To examine the effect of the boundary condition, we use the same four wind datasets mentioned above to drive the model except with a close boundary (without the Kuro- shio’s effect). As shown in Fig. 6a, the modeled circulation pattern without the Kuroshio is generally similar to that with the open boundary condition, except in the area near Luzon Strait (Fig. 5a). The result is consistent with the previous studies that showed that the SCS circulation is mainly driven by the monsoon (Liu et al. 2001; Wang et al. 2006). The TCs-induced circulation patterns with the close boundary condition (Fig. 6b–d) are also similar to those with the open boundary condition (Fig. 5b–d). All of them show that both the cyclonic gyre and the anti-cyclonic gyre are enhanced by the TCs. However, the TCs-induced cyclonic circulation in the north with the close boundary condition is slightly stronger than that with the open boundary condition (Figs. 5b and 6b), suggesting that interaction between TCs and the Kuroshio can produce a negative vorticity. Since the model includes the nonlinear terms, it will be valuable to evaluate the nonlinear effect on the TCs- induced circulation. We use the same four wind forcing datasets to drive the model excluding the nonlinear terms. As shown in Fig. 7, the Kuroshio can intrude into the SCS if the nonlinear terms are excluded, which is consistent Fig. 7 The same as Fig. 5, except for the model experiment without with previous model results (Zhang et al. 2006). Except for the nonlinear effect. a The dashed (solid) contours are for the the anti-cyclonic gyre in the northeastern SCS, the SCS thermocline depth smaller (larger) than 175 m. b–d The positive values are shaded circulation also includes a cyclonic gyre in the north and an anti-cyclonic gyre in the south. It is interesting to note that both the north cyclonic gyre and the south anti-cyclonic gyre are still enhanced by the TCs, although the nonlinear terms are excluded. This suggests that the nonlinear terms are not important to the TCs-induced circulation, in spite of its significant impact on the Kuroshio intrusion. All of the modeled results are qualitatively consistent with the Sverdrup theory. Both SCS and NWP TCs can enhance the SCS double gyres in summer. Although the number of NWP TCs is much larger than that of the SCS, the effect of NWP TCs on the summer SCS circulation gyres is weaker than that of SCS TCs. This is because the positive wind stress curl induced by NWP TCs is more northwestward than SCS TCs, and NWP TCs produce a negative wind stress curl in the northwest of Luzon Islands. The available navy daily satellite SSHA dataset is used to Fig. 8 The summer averaged SSHA (cm) induced by a NWP TCs verify the model results. Since the inertial oscillation and b SCS TCs induced by TC in SCS can last 6–8 days after the TC passage (Zhu and Li 2007), we define the difference of the SSHA in Fig. 8, both NWP TCs and SCS TCs can enhance the between 1 day before the TC passage and 7 days after the TC northern cyclonic gyre and the southern anti-cyclonic gyre. passage as the SSHA difference induced by TCs. As shown The dipole off central Vietnam is also much intensified by the 123 354 Z. Ling et al.

Fig. 9 The ratio (in %) between the TC-induced thermocline depth variance and the magnitude of the thermocline depth during summer for: a all TCs, b NWP TCs, and c SCS TCs

two types of TCs. It is also shown that SCS TCs can induce a reduced gravity model as well as observations to show that stronger double gyres than NWP TCs. All these are consis- both TCs generated in the SCS and in the NWP play an tent with the models’ results described above, giving confi- important role in the summer SCS ocean circulation. This dence in the results presented in this paper. suggests that a complete understanding of SCS ocean cir- culation requires us to consider and include the TCs’ influence. In addition, more complicated ocean general 5 Summary and discussion circulation models are needed to investigate the effect of TCs on SCS ocean circulation. This study shows that the wind stress curl induced by TCs generated either over the NWP or the SCS can affect the Acknowledgments We thank the editor and the three anonymous summer SCS ocean circulation. In summer, both NWP and reviewers for their valuable comments and suggestions to improve the quality of the paper. This study was supported by China’s National SCS TCs tend to enhance the summer circulation pattern of Basic Research Program (2007CB816004), the International Corpo- the cyclonic gyre in the northern SCS and the anti-cyclonic ration Program of China (2008DFA22230) and NSFC (Grant no. gyre in the southern SCS. However, the relative importance 40976017, 40730843). of NWP and SCS TCs in the summer SCS circulation is different. The effect of SCS TCs on summer SCS circu- lation is much larger than that of NWP TCs although the References number of SCS TCs is smaller than NWP TCs. This is because the SCS TCs-induced wind stress curl pattern is Chu PC, Lv SH, Liu WT (1999) Uncertainty of the South China Sea prediction using NSCAT and NCEP winds during tropical storm more favorable for enhancing the summer SCS circulation Ernie 1996. J Geophys Res 104:11273–11289 pattern. Generally speaking, neither the Kuroshio effect nor Cai SQ, Long XM, Wu RH, Wang SG (2008) Geographical and the nonlinear effect are sufficiently significant to affect the monthly variability of the first baroclinic Rossby radius of TCs-induced summer circulation. deformation in South China Sea. J Mar Syst 74:711–720 Hurlburt HE, Thompson JD (1980) Numerical study of loop current To evaluate the contribution of the TC-induced circu- intrusions and eddy shedding. J Phys Oceanogr 10:1611–1651 lation, we calculated the ratio between the TC-induced Lin II, Liu WT, Wu CC, Chiang JCH, Sui CH (2003) Satellite summer thermocline depth variance and the magnitude of observations of modulation of surface winds by typhoon-induced the summer thermocline depth variability, as shown in upper ocean cooling. Geophys Res Lett 30(3):1131. doi: 10.1029/2002GL015674 Fig. 9. All TCs can contribute about 0–18% of the original Liu CX, He XC (2003) The Analysis on the statistical character of circulation (Fig. 9a). The SCS and NWP contributions are QuikSCAT scatterometer winds and strong wind frequency about 0–12% and 0–6%, respectively (Fig. 9b, c). These using remote sensor data from QuikSCAT. J Trop Meteor values indicate that the TC influence is an important factor 19:107–117 (in Chinese) Liu ZY, Yang HJ, Liu QY (2001) Regional dynamics of seasonal in affecting SCS upper layer circulation and that the con- variability in the South China Sea. J Phys Oceanogr 31:272–284 tribution by TCs generated in the SCS is larger than those Metzger EJ, Hurlburt H (1996) Coupled dynamics of the South China in the NWP. Sea, Sulu Sea, and the Pacific Ocean. J Geophys Res As stated in the ‘‘Introduction’’, most SCS studies do not 101:12331–12352 Metzger EJ, Hurlburt H (2001) The nondeterministic nature of consider the effect of TC-induced wind forcing on the SCS. Kuroshio penetration and eddy shedding in the South China Sea. The present paper uses the simple Sverdrup theory and a J Phys Oceanogr 31:1712–1732

123 Different effects of tropical cyclones generated in the South China Sea 355

Qu TD (2000) Upper-layer circulation in the South China Sea. J Phys Wang G, Wang C, Huang RX (2010) Interdecadal variability of the Oceanogr 30:1450–1460 eastward current in the South China Sea associated with the Shaw PT, Chao SY, Fu LL (1999) Sea surface height variation in the summer Asian monsoon. J Clim 23:6115–6123 South China Sea from satellite altimetry. Oceanol Acta 22(1):1–17 Wrytki K (1961) Scientifc results of marine investigation of the South Su J (2004) Overview of the South China Sea circulation and its China Sea and Gulf of Thailand. NAGA report 2 influence on the coastal physical oceanography near the Pearl Xie SP, Xie Q, Wang DX, Liu WT (2003) Summer upwelling in the River Estuary. Cont Shelf Res 24:1745–1760 South China Sea and its role in regional climate variations. Sverdrup HU (1947) Wind-driven currents in a baroclinic ocean; with J Geophys Res 108(c8):3261. doi:10.1029/2003JC001876 application to the equatorial currents of the eastern Pacific. Proc Xu XZ, Qiu Z, Chen HC (1982) The general descriptions of the Natl Acad Sci USA 33:318–326 horizontal circulation in the South China Sea. In: Proceedings of Wang G, Su J, Chu PC (2003) Mesoscale eddies in the South China the 1980 symposium on hydrometeorology. Chinese Society of Sea detected from altimeter data. Geophys Res Lett 30(21). doi: Oceanology and Limnology, Science Press, Beijing, pp 137–145 10.1029/2003GL018532 (in Chinese) Wang G, Chen D, Su J (2006) Generation and life cycle of the dipole Yin X, Wang Z, Liu Y, Xu Y (2007) Ocean response to Typhoon in South China Sea summer circulation. J Geophys Res Ketsana traveling over the northwest Pacific and a numerical 111:C06002. doi:10.1029/2005JC003314 model approach. Geophys Res Lett 34:L21606. doi:10.1029/ Wang G, Su J, Ding Y, Chen D (2007) Tropical cyclones genesis over 2007GL031477 the South China Sea. J Mar Syst 68:318–326 Zhang QH, Fan HM, Qu YY (2006) Kuroshio intrusion into the South Wang G, Chen D, Su J (2008) Winter eddy genesis in the Eastern China Sea. J Hydro 18:702–713 South China Sea due to orographic wind-jets. J Phys Oceanogr Zhu DY, Li L (2007) Near inertial oscillations in shelf-break of 38:726–732 northern South China Sea after passage of typhoon Wayne. Wang G, Ling Z, Wang C (2009) Influence of tropical cyclones on J Trop Oceanogr 26(4):1–7 seasonal ocean circulation in the South China Sea. J Geophys Res 114:C10022. doi:10.1029/2009JC005302

123